mixer

Reusable, pure-Dart PCM mixing for ManyMath audio projects.

mixer_dart owns device-independent audio-graph behavior: concurrent clip voices, gain, pause/resume/seek/loop controls, speaker-layout channel mapping, sample rate conversion, summing, and S16LE output clamping. It intentionally does not open microphones or speakers. Packages such as soundboard route its output to sound_dart, microphone_dart, virtual devices, files, or network transports.

Clips are resampled with a Kaiser-windowed sinc filter by default, which low-passes below the output's Nyquist frequency so downsampling onto an 8 or 16 kHz voice bus does not alias. Pass resampling: PcmResampling.linear for cheaper linear interpolation. Up- and down-mixing assume the WAVE/SMPTE channel order for 1 to 8 channels and fold surrounds into stereo ITU-style, without the LFE.

final mixer = PcmMixer(
  format: const PcmFormat(sampleRate: 48000, channels: 2),
);
final voice = mixer.play(clip, gain: 0.8);
mixer.frames.listen(sendSomewhere);

By default a timer clocks the mix. When the output feeds a device with its own clock, such as a microphone capture stream, let that device drive it instead so the two never drift apart:

final clock = mixer.attachClock()!;
final frames = clock.pull(framesTheDeviceJustDelivered);
clock.detach(); // back to the timer

The repository uses a Dart workspace with Melos for tooling. CI is not configured yet.